Risk of leakage through wellbores:g gIs it really that high?Matteo Loizzoa, Salvatore Lombardib, Onajomo Akemua, Laurent Jammesa,

bA. Annunziatellisb

aSchlumberger Carbon Services, bUniversitá di Roma “La Sapienza”

IEAGHG 6th Wellbore Network MeetinggNoordwijk Aan Zee, The Netherlands

2010 Apr 29

© 2010 Schlumberger. All rights reserved.

An asterisk is used throughout this presentation to denote a mark of Schlumberger. Other company, product, and service names are the properties of their respective owners

2

properties of their respective owners.

Outline of the presentation

● Analogs to CO2 storage wells● Well analogs

― Major eventsWell head pressures and leak rate distribution― Well-head pressures and leak rate distribution

― Leakage pathways― Risk profile and leak rate distribution

Eff t f ti― Effect of time● Natural analogs

― Distribution and consequencesq● Containment monitoring● Regulation: US and EU● Conclusions

3

Carbon storage – well analogs

● CO2: high-pressure, light, thin fluid―Driving force, buoyancy, low friction pressure drop

● “Only when pressures are exerted on the well as during fracturing, acidizing, or fluid injection of a flooding operation do some of these [primary-cementing] failures become apparent ” [Scott & Brace 1966]failures become apparent. [Scott & Brace, 1966]

―Methane, steam● “Consideration of CO2 versus steam injection suggests the abandoned-well

blowout rate in CO2-storage fields may not be dissimilar” [Preston & Benson blowout rate in CO2 storage fields may not be dissimilar [Preston & Benson, 2008]

● Methane Lower Explosive Limit → 5%● High pressure water → water injection, hydraulic fracturing?g j y g

● Wellbore integrity―Leaks through cemented wellbores

Excluding well control incidents and tubing leaks―Excluding well control incidents and tubing leaks

4

Well analogs – major and catastrophic events

● Underground Gas Storage (UGS), depleted reservoirs and aquifers Total Events with Events with depleted reservoirs and aquifers― [Evans, 2009], worldwide – see table

● Events with wellbore integrity issues 27% of total

events casualties fatalities

O&G reservoirs

All 27 5 0

Wells 11 2 0total― Using [Papanikolau, 2006] and [Keeley,

2008] → estimate of 705,536 well-years of operation

AquifersAll 24 3 1Wells 3 0 0

Total UGS analog 14 2 0operation● 2.0 10-5 major events per well-year, 2.8 10-6 if

only casualties are considered

● Steam injection [Jordan & Benson 2008] District 4 in California

events 14 2 0

● Steam injection, [Jordan & Benson, 2008], District 4 in California― Wells in operation, thermal fields → 7.3 10-5 major events per well-year

● 9.5 10-6 for non-operational wells (7.7 times less)N f t liti― No fatalities

5

Well analogs – SCP occurrence

Large sample studies

[Marlow, 1989], survey 6.1% of c. 7,000 UGS wells in the USALeakage rate: 61% of wells leak <35 t/y; 90% of wells <200 t/y Leakage rate: 61% of wells leak <35 t/y; 90% of wells <200 t/y,

[Burgoyne et al., 1998] 11.6% of c. 30,000 wells in the Gulf of Mexico leak through casing strings

[W t & B h 2009] 9 8% f 20 000 ll i th T t A i Alb t[Watson & Bachu, 2009] 9.8% of c. 20,000 wells in the Test Area in Alberta6.3% of wells leak gas through the soil (GM) → almost 1:1 ratio to casing leaks

Smaller samples, anecdotal studies

[Watters & Sabins, 1980] 15% of 250 casing strings

[Xu & Wojtanowicz, 2001] 85% of 26 wells (Gulf of Mexico)[ u & ojta o c , 00 ] 85% o 6 e s (Gu o e co)

[Chilingar & Endres, 2004] 75% of 50 wells in Santa Fe Springs oilfield

6

Note: Sustained Casing Pressure (SCP)=Surface-Casing Vent Flow (SCVF)

Well analogs – leak rates, 1

● Risk ← loss― Little dependence on actual rate, but rather on leak hitting a “target” (exposure)Little dependence on actual rate, but rather on leak hitting a target (exposure)

● In general, leak rate data rare and less reliable than pressure● Continuous leak rate and instantaneous leak rate

― Most major leak events involved slow charging aquifer and sudden larger releases― Most major leak events involved slow charging aquifer and sudden larger releases

● [Evans, 2009] reports data on a number of UGS leaks― Leidy Field (PA 1969) depleted O&G field Leidy Field (PA, 1969), depleted O&G field

● Low-level charge of aquifer at 208 t/y (through 5 wells)● Sudden blow-out at 14,000 t/y after subsequent parting of 30 wells

― Kalle (Germany, 1999) → max 15,000 t/y over extended period, probably via 2 wells● [Araktingi et al., 1984] study on leak in Leroy UGS (WY)

― 1,900-2,300 t/y through possibly 2 wells and caprock, bubbling at surface● Tek estimate of leakage from Playa Del Rey UGS at 1,900 t/y

― 1,900 t/y =100 Mft3/y → anchoring?

7

Well analogs – leak rates, 2

● [Burgoyne et al., 1998], case history 1W ll l ki 35 000 t/ th h 9⅝” i (10 000 ft d )― Well leaking ~35,000 t/y through 9⅝” casing (10,000 ft deep)● “Very low rate […] on the order of supply gas releases […] and pilot lights”● Possible mistake → either MCF/Y or TCF/D (35-100 t/y)

[Alli 2001] d t il d i f Y l k (H t hi KA)● [Allison, 2001] detailed review of Yaggy leak (Hutchinson, KA)― Storage in salt caverns, leak possibly through hole in casing during re-drilling, 2

fatalities due to explosion & fire2 716±437 t lost possibly most of it leaked through 2 short events and relief wells ― 2,716±437 t lost, possibly most of it leaked through 2 short events and relief wells → instantaneous rate ~100,000 t/y

● [Marlow, 1989] surveyLeak rate >1 400 t/y <7% of sample

800

1000CO2 flux as a function of time

― Leak rate >1,400 t/y → <7% of sample● CO2 leaks

― [Loizzo et al., 2009], Kaniow CO2 injection wellMi ti t i f d f l 100 800 t/

200

400

600

QC

O2 [t

on/y

]

● Migration rate inferred from logs → 100-800 t/y

8

0 50 100 150 200 250 3000

CO2 flowrate vs. time (days) from Kaniow well, [Loizzo et al., 2009]

Wellbore leaks – pathways

● Four broad classes of leakage pathways:1 Placement – liquid cement → mud “channels”1. Placement – liquid cement → mud channels2. Cement setting → “chimneys”, gas and brine3. Solid cement → “microannuli”, and possibly cracks4 No cement – trivial pathway4. No cement trivial pathway

● Possible cause of brine flow into Navajo aquifer (Utah), from EPA report● Channels broadly solved by the O&G industry in the 1990’s

Sharp drop in leak rates in Alberta in 1990?― Sharp drop in leak rates in Alberta in 1990?― Gas chimneys qualitatively understood and reduced by improved practice― Technology can help drive down leaksLeak rates strongly dependent on well risk geology and drilling● Leak rates strongly dependent on well risk → geology and drilling― SCP >2x likely in the Test Area than in the whole of Alberta

● All major leaks include accumulation and transport through “imperfect collector ” h ll ifzone” → shallow aquifers

9

Well analogs – risk profile

● Risk assessment based on frequency data presented― Leaks to surface → ~10% through casings to well-

head 10% through soilEvent likelihood

(events per well per year)head, 10% through soil● Prevalence of subsurface fluid migration (e.g. SACROC)?● Leaks happen very early, but assuming well life of 10

years → frequency of 2 10-2 events/well/year● Serious events

1E-6 4E-5 2E-3 1E-1 5E+0

Light -1 -2 -3 -4 -5

(events per well per year)

Serious events– Cost around $10-50K at some point during well life, even

though they don’t result in immediate intervention― Major leaks: 5 10-5 events/well/year

● No catastrophic event recorded → assumed to be<10 6 f t liti / ll/

Serious -2 -4 -6 -8 -10

Major -3 -6 -9 -12 -15

C t t hiEven

t sev

erity

<10-6 fatalities/well/year● Considered acceptable limit for general public →

likelihood of lightning strikes (1.7 10-7)● “No [drinking water] or atmospheric endangerment by CO2

emissions from CO2 EOR projects” [Sweatman

Catastrophic -4 -8 -12 -16 -20

Multi-catastrophic -5 -10 -15 -20 -25

E

emissions from CO2 EOR projects [Sweatmanet al., 2009]

● Maximum Criticality Severity → small surface leaks (SCP/GM)

Severity causing maximum loss reference for risk

Acceptable

Demonstrate ALARP(As Low As Reasonably Practical)

Unacceptable― Severity causing maximum loss, reference for risk

management

10

Risk profile – implications

● HSE analogySt i h dli d lifti d i i ― Stepping, handling and lifting vs. driving or pressure

― Most events have relatively benign outcomes → sprained ankles or cut fingers● Medical analogygy

― Seasonal flu vs. tumors― Flu → widely spread, yet very seldom fatal

● 30 million outpatient visits every year in the US alone30 o ou pa e s s e e y yea e US a o e● Exercise reasonable prevention, see doctor if complication, vaccine every Fall

― Tumor → less common, but potentially fatal● Intensive prevention, frequent tests, immediate reaction, tolerance for false positivesIntensive prevention, frequent tests, immediate reaction, tolerance for false positives

● Acceptable risk → at most one-off awareness training, general procedures● ALARP → must evaluate risk before every exposure, specific design

― No general solution, every field/well should be optimized

11

Well analogs – SCP and leak rate distribution

100%

40%

60%

80%

lative

 freq

uency

0%

20%

0 500 1000 1500

Cumul

P ( i) di t ib ti b i h i t f Estimated leak rate (metric tons/year)

● Gas production well → mass flow rate roughly proportional to well-head pressure

Pressure (psi) distribution by occurrence in each casing type from [Burgoyne et al., 1998]

Leak rate distribution from survey, modified from [Marlow, 1989]

p g y p p p― Q~pWH

3 if flow in fracture instead of permeability● [Araktingi et al., 1984] study on leak in Leroy UGS, leakage fitted to Q~pBH

2

― Very narrow pressure range (60 psi around 1,800 psi) and exponent 2 assumed● [Burgoyne et al., 1998], case history 1

― 4,400 psi SCP, consistent with gradient used in order of magnitude calculation12

Well analogs – leak rate distribution

● Cumulative distribution of leak rate and well head pressure approx linear on a well-head pressure approx. linear on a log-log scale― Similar slope of leak rate and third power of

pWH → possible effect of flow in fracture?10

-1

100

Cumulative distribution (cdf) of Sustained Casing Pressure parameters, normalized by their median

WH― Power law exponent ~-1.5

● Leaks 10 times bigger are 3 times less likely

● Gutenberg-Richter law for earthquakes → 10

-2

10

ve p

roba

bilit

y P

(x>X

)

Gutenberg Richter law for earthquakes 10 times bigger, 10 times less likely (exponent -2)

● Even with standard O&G technology, <2 wells in 1000 will leak >10 000 t/y

10-3

Cum

ulat

iv

Leak rate (t/y) from [Marlow, 1989], divided by 18.6Linear regression of rate (slope=-0.48)Well-head pressure (psi), from [Burgoyne et al., 1998], divided by 497Linear regression (slope=-1.89)

Linear regression of pWH3 (slope=-0.63)

wells in 1000 will leak >10,000 t/y― Major events → 10,000-100,000 t/y― Would take 1,500 years to accumulate CO2

released from Lake Nyos

100

101

102

10-4

Adimensional SCP parameter

y

13

CO2 leaks – effect of time

● CementS i b i l f l i i―Strong reaction → carbonation, reversal of calcination

―Change in mechanical and transport properties, formation of fronts―Possibly degradation, maybe pathway healing under some conditions

● Need for modeling, representative transport/reaction experiments, detailed observations

● Steel―Strong reaction → uniform “sweet” corrosion―Very fast degradation (~1 year) in the absence of protective layers or

Corrosion Resistant Alloysy―Very limited research

● Cement effective protective layer, what about carbonated cement?― If cement is present, risk may not increase. If cement is absent, risk will If cement is present, risk may not increase. If cement is absent, risk will

increase quickly

14

Natural analogs – distribution

● Gas migration studied in Italy for >25 years― From volcanic & geothermal areas and oil &

gas fields● Total area covered by CO2 soil gas surveys

000 2→ 4,000 km2

― CO2 in soil air above 20% over 0.15% of area― Above 60% over 0.025% of area (1 km2)― 50% of samples with concentration <2%

● Total area covered by He-4 soil gas surveys → 12,000 km2

Surveyed areas

. CO2 emanationsHistogram of CO2

― He-4 >500 ppb above atmospheric content over 2.5% of area

― He-4 >2000 ppb above atmospheric content over 0 03% of area (4 km2) 50

500

5000

No

of o

bs

over 0.03% of area (4 km2)

150.1 7 14 21 28 35 42 49 56 63 70 77 84 91 98

CO2 (%, v/v)

5

Natural analogs vents – CO2 flux from Latera (Italy)

CO2 fluxsite

CO2 flux (g/d)

1 6599818

3 locations account for 97 5% of CO

2 87162

3 145784

4 6471988 97.5% of CO2flux

4 6471988

5 69159

6 35547

7 199989

8 7544034

total 21153480 ~7,700 t/y

Natural analogs – consequences

● Natural leaks concentrated over small area

● Very little disruption on ecosystem or human activityhuman activity―Photo from Latera caldera showing

effect of CO2 leak on vegetation and prints of sheep from nearby flocks prints of sheep from nearby flocks possibly drinking at mineral water puddle

―Kaolin quarry (with disused mineshaft) q y ( )around actively degassing fault

―Total flux from Latera caldera→ ~0.01-0.1 Mt/y

CO2 bubbling inLatera caldera (Italy) with sheep prints around the puddle, courtesy of B. Lecampion

17

y

Containment monitoring

● Part of the site monitoring plan● Verification monitoring → prevention―Ensure integrity of the vertical barriers is

preservedMinimal Costs:Risk Management:preserved

● Update risk profile → prevention measures

● Assurance monitoring → mitigationAbid L &

For each technique, For the overall plan over time

Sit & t h i l

Loss of performance inInjectivity, Capacity, Containment

―Detection of migration and leaks―Quantification of leaks

● Emission credits

Abide Laws & Regulations:

Storage safety, Credit allocation, Environmental Impact

Site & technicalconstraints:

Deployment restrictions,Tools sensitivity

● Leak detection threshold―Monitoring of HSE consequences of a leak

● Air potable aquifers mineral resources● Air, potable aquifers, mineral resources

18

US and EU regulation

US EPA 40 CFR parts 144 and 146 EU directive 2009/31/EC

Application for storage permit must include delineation of area of reviewand a “corrective action plan”

Application for storage permit to include “proposed correctivemeasures plan”and a corrective action plan measures plan

Identify all artificial penetrations in the area of review (AoR) In the event of leakages, the operator must immediately notify thecompetent authority, and take the necessary corrective measures,including measures related to the protection of human health.

Compile, tabulate, and review available information on each well in theAoR that penetrates into the confining system,

The corrective measures shall be taken as a minimum on thebasis of the corrective measures plan

Identify the wells that need corrective action to prevent the movement ofCO2 or other fluids into or between USDWs.

The competent authority may at any time require the operator totake the necessary corrective measures…additional to or differentfrom those laid out in the corrective measures plan.

Perform corrective action to address deficiencies in any wells regardlessPerform corrective action to address deficiencies in any wells, regardlessof ownership, that are identified as potential conduits for fluid movementinto USDWs.

In the event that an owner or operator cannot perform the appropriatecorrective action the Director would have discretion to modify or deny the

19

corrective action, the Director would have discretion to modify or deny thepermit application. Corrective action could be performed prior to injectionor on a phased basis

Conclusions

● Cemented wellbores very effective at controlling leak rates― Statistical data about analogs (UGS, steam injection) suggest a rate of ~5 10-5

j t / ll/ d f t litimajor events/well/year, and no fatalities― Majority of leaks ~100-1,000 t/y, with at most 2 in 1000 >10,000 t/y

● Natural analogs suggest very low or no impactS ll l k b f t f i j ti ll● Small leaks may be frequent for injection wells― Possibly 20% of all wells, with about half not showing up at the well-head― Risk also depends on geology

● Technology can help minimize leak occurrence― Prevention

● Low severity → mitigation can be effective to manage risky g g― Monitor leaks to predict evolution and plan intervention without costly shut-downs

● “Approval for departure” from 30 CFR 250.517 in the US Outer Continental Shelf● Undetected leaks can accumulate and possibly degrade pathways over a long time

● Approach should evolve from “no leak” to “no damage”

20